Moving Buffalo Farming beyond Traditional Areas: Performances of Animals, and Quality of Mozzarella and Forages

Abstract

An observational case study was designed to highlight issues associated with a possible expansion of dairy buffalo (Bubalus bubalis) farming outside the traditional coastal plains of southern Italy. Twenty pregnant buffaloes were transferred to a hilly inland farm. After calving, production and reproduction data were collected monthly throughout lactation. From 4 to 6 months of lactation, buffaloes were enrolled in a feeding trial to evaluate the effects of locally grown forages (maize silage vs. hay) on milk production and in vivo digestibility. Sensory properties of mozzarella cheese produced at a local dairy were also evaluated. No obvious effects of diet were found. Compared to the data recorded in the previous lactation completed in the farm of origin, milk yield was reduced by 37.2%, and milk protein by 6.1%, whereas milk fat improved (+4.5%). A lower pregnancy rate (−13.3%), increased days open (+122%), and a prolonged intercalving period (+26.9%) were also observed. Lactation length was shorter than the standard value of 270 d. The results showed that peculiar reproductive characteristics, lower environmental temperatures, and the specificity of the mozzarella production process are the main problems to be addressed in an expansion of buffalo farming outside traditional areas.

1. Introduction

Italian buffalo (Bubalus bubalis, Mediterranean type) farming is a traditional enterprise almost exclusively devoted to producing mozzarella cheese [1]. Currently, roughly 75% of the European 420,000 dairy buffalo heads are raised in specific coastal plains of Campania, a region of southern Italy [2,3,4]. The growing demand for buffalo mozzarella in the last few decades has driven a rapid intensification of the sector in terms of increased use of off-farm inputs and implementation of genetic improvement techniques [5,6]. The shift from extensive to intensive modes of production has increased the stocking rate and then the environmental impact of buffalo farming in terms of worsening air and water quality, as well as animal welfare and health, even facilitating occasional animal disease outbreaks [7,8,9,10]. Moving buffalo farming from the traditional coastal plains to inland hilly zones could mitigate the issues related to buffalo farming along with revitalizing the economy and contrasting the depopulation of more marginal zones [11,12,13]. Nevertheless, this expansion should consider the different environmental and infrastructural conditions of hilly and inland areas compared to lowland farms. Buffaloes originate and are mainly distributed in tropic and sub-tropic environments, so that they present anatomic and physiological mechanisms, for instance the sparse hairs and the numerous melanin pigments on epidermis, able to counter hot and humid climate conditions rather than the cooler conditions of internal areas [14,15,16]. In addition, some specific features of buffalo reproduction, such as seasonality and estrus behavior, can reduce the reproductive efficiency and then profitability of farms [17,18].

This study aimed to assess the possibility of expanding buffalo farming beyond the traditional coastal areas and to highlight related issues, and an observational case study was designed by moving a group of lactating buffaloes in an internal hilly area. We also assessed the yield performance of the buffaloes fed with locally produced forages (hays vs. maize silage), based on different scenarios for irrigation water availability and farm size. The objectives of this research can be summarized as follows: (i) assess the adaptability of buffaloes to the conditions of internal hilly areas by surveying the productive and reproductive performance throughout the whole lactation; (ii) compare the yield performance of buffaloes fed diets based on hay or maize silage; (iii) evaluate the sensory characteristics of locally produced mozzarella.

2. Materials and Methods

2.1. Study Site

The study site was a dairy cattle farm (55 ha Utilized Agriculture Area, 120 cows) sited in an internal hilly area of Campania Region (41°17′ N, 15°02′ E, 540 m a.s.l.). The area is bounded by the pre-Apennine mountains, with a Mediterranean sub-continental climate (Köppen zone Csb) characterized by dry and warm summers. The rainfall pattern is very irregular and strongly influenced by the interaction between the wet air masses and orography [19]. The average annual rainfall ranges from 800 to 1100 mm, with the maximum monthly precipitation during November and December, and minimum values during July and August. The temperatures average 4 °C in winter and 20 °C in summer [20,21].

2.2. Productive and Reproductive Performances

In early summer 20 pluriparous (age 6.0 ± 1.9 years, parity 3.2 ± 1.8), pregnant (on average 7.7 months of gestation) dry buffaloes were moved from farm of origin (41°05′ N, 14°06′ E, 13 m a.s.l) and introduced in the experimental farm, where they were kept together in an open barn equipped with external paddock, water bowls and manger until calving that took place in the farm. The observation period started about two months after the relocation, in July, at the time of delivery, and covered the whole lactation period (i.e., until dry off time). The animals were fed the same maize silage-based total mixed ration (M-TMR) formulated for a milk yield of 8.0 kg/d which was modified at the 6th month of lactation for meeting the lower needs [22,23]. After a voluntary waiting period of 60 d from parturition, on the basis of visual observation of the estrous signals, the buffaloes were artificially inseminated by intrauterine deposition of frozen-thawed semen from two proven bulls. Pregnancy diagnosis was assessed by rectal palpations at 30-day intervals, and days open (d from calving to pregnancy diagnosis), calving interval, and conception rate (percentages of cows pregnant) were calculated. Every month, milk from each cow was measured and sampled to assess for fat, protein, lactose (Milkoscan 605, Foss, Hillerød, Denmark), and somatic cells count (SCC; Fossomatic 90, Foss Electric, Hillerød, Denmark). Mozzarella cheese yield was estimated based on the milk chemical composition by using the equation of Altiero et al. [24].

Climatic data (daily air temperature, relative humidity, and rainfall) from the experimental farm and the farm of origin were collected.

2.3. Feeding Trial

At the 4th month of lactation, the cows were weighted, scored for body condition score (BCS) [25], and randomly split into two groups homogenous for milk yield and quality. One group continued to be fed the M-TMR whereas the other was fed an isonitrogenous and isoenergetic ration in which the maize silage was substitute by ryegrass hay (H-TMR) (

Table 1. Ingredients (kg/d as-fed) and chemical composition (% of dry matter, DM, if not otherwise stated) of the total mixed rations fed to the study buffaloes.

	M-TMR	H-TMR
Ingredients
Maize silage	15.4	-
Alfalfa hay	3.1	4.0
Ryegrass hay	2.0	5.0
Concentrates	6.3	7.0
Hydrogenated fat 1	-	0.3
Chemical composition
Dry matter (kg/d)	14.7	14.4
Ash	8.2	9.5
Crude protein	15.2	15.5
Ether extract	3.1	4.1
Starch	25.4	18.0
Neutral detergent fiber	42.7	45.5
Acid detergent fiber	27.8	31.5
Acid detergent lignin	6.3	7.4
NEL (UFL/kg DM)	0.85	0.86

¹ Supplement containing 4% hydrogenated palm oil (Hidrofat© Nutrición Internacional, S.L., Madrid, Spain). M-TMR, maize silage based total mixed ration; H-TMR, hay-based total mixed ration; NE_L, net energy of lactation; UFL, Unité Fourragère Lait: 1 UFL = 7.11 MJ/kg of NE_L; DM, dry matter.

). The trial lasted 8 weeks, after 10 d of adaptation. The groups were housed in two adjacent free stall barns and handled in similar way in terms of feeding and management.

Dry matter intake (DMI) was measured every week on a group basis along with milk and rations sampling. The TMRs were analyzed according to AOAC [26] for dry matter (DM), ash, fat (Soxhlet apparatus), and crude protein (CP, Kjeldahl method). Neutral detergent fiber including residual ash (aNDF), acid detergent fiber (ADF), and acid detergent lignin (ADL) were determined according to Van Soest et al. [27]. Starch content was assessed by a Polax-2l polarimeter (Atago Co., Ltd., Tokyo, Japan) in a 200 mm long observation tube [28]. Net Energy for Lactation (NEL) was calculated based on chemical composition [29]. At the 7th week of the experimental period, the total tract in vivo digestibility was evaluated by using acid insoluble ash (AIA) as intrinsic marker [30] according to the procedure described by Vander Pol [31] and Masucci et al. [32]. Briefly, for three consecutive days, fecal grab samples were collected from the rectum of each animal at 0900 h–1300 h and 1700 h. Samples of the rations were also collected and DMI determined. The samples were dried to determine partial DM, composited per animal (fecal samples), and analyzed for CP, NDF, ADF, and AIA content.

2.4. Mozzarella Cheese Production

Because traditional mozzarella cheese is an artisanal product subject to variation in relation to cheesemaking conditions, the consistency of the sensory properties of mozzarella on different days of production was tested. It was not possible to produce mozzarella separately for the M-TMR and H-TMR groups, as no cheesemaking equipment was available to process small amounts of milk in separate tanks at the same time. Therefore, at the end of the feeding trial (sixth month of lactation) the buffaloes were reunited and fed the same maize silage-based diet. After two weeks of adaptation, the mozzarella production process was carried out on two different days (D1 and D2), one week apart, by the same staff from a local dairy. Milk was collected at evening and morning milking, transported to the dairy, and manufactured for mozzarella production according to the traditional procedure [33].

2.5. Chemical and Sensory Analyses

Before the mozzarella manufacturing process, milk was sampled to determine chemical composition and titratable acidity [26]. Mozzarella samples from the D1 and D2 batches of production were separately analyzed for macro-component and quantitative descriptive sensory analyses (QDA) after 24 and 48 h from manufacturing. Overnight, mozzarella samples were kept at 10 °C and allowed to equilibrate at room temperature (22–23 °C) for at least 4 h before analysis. Chemical composition was determined on 3 samples/batch, each consisting of 4 blended pieces of mozzarella. Moisture was quantified by oven drying while fat and protein were determined according to the Gerber and Kjeldahl methods, respectively [26].

To perform a quantitative descriptive analysis (QDA) [34], 15 subjects were recruited among regular eaters of buffalo mozzarella cheese (defined as consuming this cheese at least once a week). In total, 10 panelists (6 females and 4 males, mean age 29 year) were selected by assessing their capacity to identify the 4 basic tastes (sourness, sweetness, bitterness, and saltiness), according to the ISO recommendations [35]. During a preliminary phase, the panelists were asked to taste buffalo mozzarella samples and, on the basis of available literature [34,36], they developed and agreed on a 19-attribute consensus list (

Table 2. List of attributes and reference samples used by the 10-member trained panel for Buffalo mozzarella cheese sensory profiling ¹.

Attribute	Reference Sample		Definition
	Lower Anchor	Upper Anchor
Appearance
Color	7.5Y 9/2 (ivory)	N 9.5 (white)	Overall intensity of white color (from ivory to white)
Brightness	Ripened scamorza cheese	Buffalo mozzarella cheese	Shininess or glossiness of surface
Stringy appearance	Ripened scamorza cheese	Nodino cheese	Typical filamentous or fibrillar structure of milk casein after hot water stretching
Whey loss	Ripened scamorza cheese	Buffalo mozzarella cheese	Amount of whey visible on the surface of the cheese sample after cutting
Skin thickness	Mozzarella cheese	Burrata cheese	Skin thickness of the buffalo mozzarella cheese evaluable after cutting
Odor/Flavor
Overall Odor	Ricotta cheese	Buffalo mozzarella cheese	Level of overall odor
Overall Flavor	Ricotta cheese	Buffalo mozzarella cheese	Level of overall flavor
Milk	Mixture of water (80%) and milk (20%)	Whole milk	Odor/flavor arising from milk at room temperature
Cream	Mixture of water (80%) and cream (20%)	Whole cream	Odor/flavor arising from cream at room temperature
Yogurt	Whole milk	Plain whole yogurt
Taste
Saltiness	1.5 mL stock solution 100 mL−1	3 mL stock solution 100 mL−1	Fundamental taste associated with sodium chloride
Sweetness	8 mL stock solution 100 mL−1	20 mL stock solution 100 mL−1	Fundamental taste associated with sucrose
Bitterness	4 mL stock solution 100 mL−1	8 mL stock solution 100 mL −1	Fundamental taste associated with quinine
Sourness	8 mL stock solution 100 mL−1	16 mL stock solution 100 mL−1	Fundamental taste associated with citric acid
Texture
Tenderness	Ripened provolone cheese	20 g round size mozzarella cheese	Minimum force required to chew cheese sample:the lower the force the higher the tenderness
Screechiness	Asiago cheese	Nodino cheese	Mouth sensation describing the typical filamentous or fibrillary texture of milk casein after hot water stretching
Elasticity	Parmesan cheese seasoned 36 mo	Emmental cheese	Degree to which the original shape of a product is restored after compression between the teeth
Moisture	Ripened scamorza cheese	Ricotta cheese	Moisture released by the product in the mouth during early mastication
Oiliness	Ricotta cheese	80 g butter mixed with 40 g ricotta cheese	Amount of oily/fatty feeling in the mouth during chewing

¹ Color definitions as in Munsell Book of Color (X Rite color. Europe GmbH). Stock solutions for sweetness, sourness, saltiness, and bitterness were, respectively: 50 g sucrose 250 mL⁻¹ solution; 2.5 g citric acid 250 mL⁻¹ solution; 25 g sodium chloride 250 mL⁻¹ solution; 51.25 g quinine hydrochloride 250 mL⁻¹ solution.

) concerning appearance (5), odor/flavor (5), taste (4), and texture (5).

Subsequently, assessors were trained with a reference frame, using specific products to each identified attribute (Table 2). During the panel training, at least two points of the scale were anchored to the reference material. In particular, they were repeatedly exposed to the reference samples (3 times) while overtly showing the corresponding intensity levels. Subsequently, panelists re-assessed the two levels of intensity of each attribute in blind conditions.

Then, a QDA was conducted in sensory booths [35,37], to assess the sensory profile of the products. The tests started at about 10.00 h and panelists evaluated two of 50 g buffalo mozzarella cheese at a serving temperature of 13 °C. To mask appearance differences in the sample, assessors evaluated the first cheese under red light for odor/flavor, taste, and texture attributes, and the second cheese under white fluorescent lighting, for appearance attributes. The panelists evaluated 3 replications of each product, at 24 h and 48 h from mozzarella manufacturing. The interval between consecutive samples was roughly 10 min and panelists drank a sip of water after each sample.

As QDA is based on the use of a 100 mm unstructured intensity linear scale, with anchor points at each end (0 = absent and 100 = very intense), panelists were trained to identify the intensity ranges for low, medium, and high intensity [36].

2.6. Statistical Analysis

A descriptive statistic was used for productive (milk yield and quality over the whole lactation) and reproductive performances. Data from the feeding trial were analyzed using SAS statistical package (version 6.09, SAS Institute, Cary, NC, USA). Milk yield and quality of the TMR-M and TMR-H groups underwent analysis of variance (ANOVA) for repeated measures (mix proc) with diet (TMR-M and TMR-H) as a non-repeated factor and sampling time and diet × sampling time as repeated factors. The cow variance was considered as random and as the error term to test the main effect of the diet. Dry matter intake, and animals’ BCS were analyzed by ANOVA (GLM procedure) to determine the fixed effects of the dietary treatment (TMR-M and TMR-H). Chemical composition of mozzarella cheese was analyzed by one-way ANOVA (GLM procedure) to determine the fixed effects of time of conservation (24 and 48 h). Sensory profile data were subjected to ANOVA with assessor (10), replication (3), storage time (24 and 48 h), day of production (1 and 2), and the interaction as factors. Data are reported as least square means (LSM) and standard errors of means (SEM). Statistical significance was declared at p < 0.05 and tendencies discussed at p < 0.10.

3. Results

3.1. Climatic Data

Data retrieved from the climatical stations are summarized in

Table 3. Seasonal climatic data (mean ± standard deviation) of the experimental farm and origin farm during the experimental period.

Item	Summer 1	Fall 2	Winter 3	Spring 4
Study site
AT 5, °C	20.1 ± 1.9	12.4 ± 4.9	2.9 ± 1.4	8.1 ± 3.2
RH 6, %	74.1 ± 3.0	83.7 ± 7.1	86.5 ± 4.6	82.0 ± 6.6
R 7, mm	26.2 ± 3.7	70.5 ± 30.7	59.0 ± 23.3	45.6 ± 26.3
Farm of origin
AT 5, °C	26.1 ± 1.6	20.4 ± 3.4	10.0 ± 3.1	13.5 ± 2.3
RH 6, %	58.8 ± 3.1	72.7 ± 4.9	72.7 ± 2.7	66.8 ± 6.6
R 7, mm	7.4 ± 5.4	32.4 ± 2.9	34.4 ± 15.7	32.9 ± 15.7

¹ Summer = June, July, and August; ² Fall = September, October, and November; ³ Winter = December, January, and February; ⁴ Spring = March, April, and May. ⁵ AT, daily air temperature; ⁶ AR, daily relative humidity; ⁷ R, monthly rainfall.

.

The study area was steadily colder (−7 °C per day), wetter (+14% of daily relative humidity), and rainier (+23.5 mm of rain per month) than the coastal area from which the animals were moved. The largest seasonal variations were in fall, when the daily temperatures and total rainfall differed by about 8 °C and 114 mm, respectively.

3.2. Maize-Based vs. Hay-Based Rations

The main differences between the chemical and nutritive characteristics of TMRs were for the contents in starch (higher in M-TMR), and NDF and ADF (higher of H-TMR), while the energy concentrations were quite close between the two diets (Table 1). The effects of dietary treatment on DMI and milk production are shown in

Table 4. Dry matter intake, body condition score, milk yield, and quality (LSM) of buffaloes fed total mixed ration based on maize silage (M-TMR) and hay (H-TMR).

Item	Diets		SEM	p Value
	M-TMR	H-TMR		D	T	D × T
DMI, kg/d	14.70	14.10	0.19	0.0357	-	-
BCS	7.07	7.06	0.26	0.9579	-	-
Milk yield, kg/d	6.09	6.69	0.74	0.5639	<0.0001	0.3500
Fat, %	9.05	8.76	0.21	0.3406	<0.0001	0.9347
Protein, %	4.61	4.60	0.05	0.9093	<0.0001	0.9959
Lactose, %	4.65	4.58	0.07	0.4555	<0.0001	0.5292
SCC, n. cells/mL	742.150	903.830	194.56	0.5603	0.2777	0.9475

M-TMR, maize silage based total mixed ration; H-TMR, hay-based total mixed ration; SEM, standard error of mean; D, diet; T, time; DMI, dry matter intake; BCS, body condition score.

. The M-TMR group showed a slightly but significantly (p < 0.05) higher DMI, whereas milk yield, milk fat, and protein were rather similar between the dietary groups along with SCC and theorical mozzarella cheese yield.

The apparent total tract digestibility as influenced by the dietary treatments is in

Table 5. In vivo digestibility coefficients (LSM) of buffaloes fed total mixed ration based on maize silage (M-TMR) and hay (H-TMR).

Item	Diets		SEM	p Value
	M-TMR	H-TMR
Dry matter	76.26	75.65	0.91	0.6244
Organic matter	77.97	77.6	0.91	0.7661
Crude protein	76.25	77.47	0.89	0.323
Neutral detergent fiber	67.11	70.56	1.23	0.0527
Acid detergent fiber	63.26	66.89	1.38	0.0661

M-TMR, maize silage based total mixed ration; H-TMR, hay-based total mixed ration; SEM, standard error of mean.

. No significant differences were detected between the groups, albeit NDF and ADF digestibility tended to be better in the H-TMR group (p < 0.0527 and p < 0.0661, respectively).

3.3. Productive and Reproductive Performances

Since the lack of statistically significant differences between the two feeding groups, milk yield and quality, over the whole lactation were calculated regardless the diet (

Table 6. Productive and reproductive performances (mean $\pm$ standard deviation) of lactating buffaloes.

Item	Present Study	Previous Lactation	Δ 1 (%)
Milk yield, kg/lactation	1546 ± 622	2463 ± 632	−37.2
Milk fat, %	8.64 ± 0.83	8.27 ± 0.28	+4.5
Milk protein, %	4.43 ± 0.32	4.72 ± 0.07	−6.1
Days open, day	189 ± 112	85± 32	+122.3
Lactation length, days	256 ± 80	314 ± 27	−18.5
Calving interval, day	499 ± 112	393 ± 32	+26.9
Pregnancy rate 2, %	65	75	−13.3

¹ Δ, calculated as percentage difference between the means recording during the study and in previous lactation completed in a plain farm. ² Calculates on a group basis as the number of cows confirmed pregnant divided by the total number of cows × 100.

).

Compared to the data recorded in the previous lactation completed in the farm of origin, milk yield was reduced by 37.2% and milk protein by 6.1%, whereas milk fat improved (+4.5%) because of the lower milk quantity. In addition, milk yield was well below the average values reported for buffalo cows bred in Campania [38,39,40,41]. Lactation length was shorter than the standard value of 270 d (Table 6). Compared to the data available for buffaloes farmed in similar conditions and seasons [42,43,44,45], and the data recorded in the previous lactation, a lower pregnancy rate (−13.3%), increased days open (+122%), and a prolonged intercalving period (+26.9%) were observed (Table 6).

3.4. Mozzarella Cheese Quality

Milk chemical composition (fat 8.61 vs. 8.76%, protein 4.79 vs. 4.81%, lactose 4.88 vs. 5.01%, respectively, for D1 and D2), titratable acidity (7.15 vs. 7.13 °SH/100 mL), as well as mozzarella yield (23.7 vs. 22.5%) were rather similar across two cheesemaking days.

By contrast, the chemical composition of the two batches of mozzarella varied over the two days of production. In fact, the D1 mozzarella had significantly higher moisture (p < 0.01), protein (p < 0.05), and fat (p < 0.01) contents than D2 (

Table 7. Chemical composition and sensory profile of mozzarella cheese (LSM) affected by cheesemaking day and conservation time.

	Cheesemaking Day		Storage Time (h)		SEM	p Value
	D1	D2	24	48		Day	ST	Day × ST
Chemical composition
Moisture	54.17	51.01	52.53	52.65	0.41	0.0006	0.8398	0.5998
Fat	28.97	25.32	27.02	27.27	0.51	0.001	0.7385	0.9111
Protein	14.77	14.35	14.55	14.57	0.09	0.0136	0.9029	0.7153
Appearance
Color	34.8	26.2	34.0	27.0	2.99	0.0457	0.1035	0.0791
Brightness	46.6	39.4	46.1	39.8	2.83	0.0753	0.1204	0.7259
Whey loss	34.1	19.1	28.5	24.7	2.61	0.0001	0.3088	0.4597
Stringy appearance	35.0	25.5	31.0	29.4	2.45	0.008	0.6513	0.2985
Skin thickness	33.0	23.0	26.9	29.2	2.40	0.0044	0.1115	0.2659
Odor/Flavor
Overall odor	46.5	46.2	45.0	45.0	2.06	0.9132	0.3677	0.6966
Yogurt odor	34.5	35.3	34.7	35.1	2.31	0.8092	0.9189	0.4872
Overall flavor	49.8	41.0	46.3	44.4	2.22	0.0063	0.5428	0.7856
Milk flavor	43.5	37.1	41.0	39.6	2.02	0.0285	0.6228	0.1128
Cream flavor	39.6	31.0	35.7	34.8	2.12	0.0056	0.7712	0.745
Yogurt flavor	25.5	22.6	2.1	2.1	2.09	0.3457	0.3181	0.1499
Taste
Saltiness	39.0	27.3	30.5	35.8	1.99	<0.0001	0.0687	0.1147
Sweetness	17.8	25.5	21.7	21.6	1.98	0.0073	0.9512	0.1251
Bitterness	9.3	8.5	9.8	8.0	1.26	0.6619	0.3218	0.8313
Sourness	35.9	17.8	29.1	24.6	2.15	<0.0001	0.1477	0.139
Texture
Tenderness	43.1	21.0	33.8	30.2	2.29	<0.0001	0.2698	0.1959
Screechiness	30.4	46.3	41.0	35.8	2.86	0.0002	0.2068	0.1813
Elasticity	37.7	44.0	38.1	43.6	2.58	0.1072	0.1617	0.1775
Moisture	54.0	21.1	37.3	37.8	2.58	<0.0001	0.8814	0.1777
Oiliness	39.0	29.4	32.3	36.1	2.39	0.0062	0.2741	0.8426

D1, first day of mozzarella cheese production; D2, second day of mozzarella cheese production; SEM, standard error of mean; D, cheesemaking day; T, storage time.

).

As for sensory analyses, no significant product x replication or product x assessor interactions were detected in preliminary ANOVA, suggesting the efficacy of the training program and of the reference frame developed in this study, both allowing to reach high reliability of the panel (i.e., products were not evaluated differently in different replications or by different assessors).

Apart from brightness (Appearance), overall odor, yoghurt odor and yoghurt flavor (Odor/Flavor), bitterness (Taste), and elasticity (Texture), the sensory attributes of mozzarella cheese produced in the two days significantly differed. Mozzarella D1 showed higher intensities perceived of the attributes related to appearance (i.e., color p < 0.05, whey releasing p < 0.001, stringy appearance, skin thickness p < 0.01), and flavor (milk flavor p < 0.05, overall flavor, cream flavor p < 0.01). Moreover, D1 was less sweet (p < 0.01), but sourer (p < 0.001) and saltier (p < 0.01). Among the texture attributes, oiliness, tenderness (p < 0.001), and moisture (p < 0.01) rated higher in D1, and accordingly, lower intensity was perceived for screechiness (p < 0.01) (Table 7).

Storage time did not affect either chemical composition or sensorial parameters. The typical sensory defect of mozzarella due to storage, such as outer skin adherence (22.52 ± 1.82 vs. 26.07 ± 1.75, p = 0.1926, respectively, for 24 and 48 h of storage time), was not significantly increased.

4. Discussion

No great effects were observed by using a hay-based diet instead of maize, as is usually the case on traditional lowland buffalo farms. The lower NDF content of the maize-based diet may have led to the higher DMI observed for M-TMR group since it is the main determinant of the rumen fill and then of the ingestion capacity in ruminants [46]. However, the higher DMI did not influence milk yield and quality and with them mozzarella cheese yield because of the similar milk protein and fat contents. Since the level of fibrous fractions in rations can greatly influence digestibility [47,48,49], a better digestibility of the less fibrous M-TMR diet would have been expected. However, the higher starch levels of this diet have likely lowered rumen pH and so impaired the activity of cellulolytic microorganisms [50]. Anyway, the tendency to a lower fiber digestibility observed for the M-TMR diet had no obvious effects either on the milk yield or on the animals’ BCS. Overall, the hay-based diet has yielded similar results to that based on maize silage, but its tendentially higher cost [51] makes it unprofitable if maize silage is available.

No complications were observed at calving time and the neonatal deaths (3 calves) were within the usual range for buffalo farming [32,52], indicating that cows’ handling and calf care practices were appropriate. An undesirable result, however, is the rather low milk production largely determined by the short duration of lactation, which does not seem to be due to feeding errors, since correct BCS and DMI and good milk production were observed during the feeding trial. Nevertheless, yield losses induced by environmental change and weather distress cannot be excluded. In this regard, after a sudden drop (−3 °C) in the minimum temperature, buffaloes at the early and intermediate stages of lactation reduced milk production up to 20% for several days after the event, thus indicating that low temperatures can have a cumulative effect [53].

The poor reproductive performances do not find an easy explanation. As buffalo is a short-day breeder and the cows calve in late summer, an early postpartum resumption of ovarian activity and a better reproduction efficiency would have been expected [54]. The shortness of the sexual receptivity period combined with the difficulties in estrus detection of buffaloes may have led to improper artificial insemination timing [54,55]. Indeed, the buffalo estrus phase averages 20 h with high incidences of short (<12 h) and medium (13–24 h) durations [56,57]. In addition, the visual observations for estrus detection are less efficient in buffaloes since, compared to cattle, the typical signs and behaviors of estrus as frequent urination, temporary teat engorgement, vulvar edema, and vocalizations, restlessness, tail raising are extremely weak and poorly represented, and many cows show estrus signs in the late-night and early morning hours [58]. The poor reproduction efficiency may have been amplified by weather-induced stress. According to Zicarelli [51,59], buffalo cows are prone to change their reproductive pattern when exposed to sudden climatic variation, becoming acyclic when cold wind and heavy rain associated with thermal drops occur. Overall, failure of instrumental insemination and cooler environmental temperatures are likely to account for the poor production and reproductive performance, while the BCS at the end of the observation period and the feeding trial data do not indicate errors in diet or feeding. In any case, these problems could be overcome by adopting insemination protocols for buffaloes and sheltering the animals from cool environments.

Both chemical and sensory characteristics of mozzarella cheese varied widely over the two days of production and appear to be closely related to each other. The higher moisture content of D1 mozzarella might explain the higher intensities perceived for whey releasing and moisture while the higher rate for cream and milk flavor, and tenderness may be ascribed to higher fat content of these samples, as assumed by Stevens and Shah [60]. The sensory quality of a fresh cheese depends on several factors linked to both the characteristics of the raw milk and the cheese-making technology [61]. The differences both in the chemical composition and in the sensory profile of the mozzarella samples seem to indicate a non-uniformity of the D1 and D2 cheesemaking processes. Factors as natural milk treatment, whey starter culture, curd pH and temperature, and management of the stretching phase can greatly influence the composition of the mozzarella, as they impact on the amount of moisture and fat trapped in the typical fibrous texture of mozzarella, as well as attributes such as structure, the stringy appearance, and skin thickness [62,63]. In addition, skin thickness seems to be also related the preservative liquid composition, and, in particular, to higher citric acid concentration [64]. These results indicate the need for adequate training of local cheesemakers on the mozzarella cheese-making process, as any perceived reduction in sensory characteristics typical of traditional dairy products may not be accepted by consumers [65].

As for the storage time effect, mozzarella is a high perishability cheese due to the high moisture content and water activity, with mass transfer between the cheese matrix and its serum phase. Probably, the storage times considered were rather short to affect many sensory attributes. In fact, other authors [66] found significant changes in volatile profile of traditional mozzarella cheese and lactose free mozzarella after 13 days and 8 days of storage, respectively; these modifications resulted from amino acid and fatty acid metabolism occurred in the samples and caused a sensory decay of positive descriptors associated with fresh cheese products and higher intensities of negative attributes, such as bitter taste, associated with release of bitter tasting peptides due to the proteolytic activity of spoilage microorganisms [67]. Moreover, a decrease in sensory hardness was significantly observed by Alinovi et al. [68] after 7 d of refrigerated storage and they related it to casein hydrolysis.

5. Conclusions

During the first year of farming in a cooler inland hilly area, multiparous buffaloes showed a reduction in milk production by more than one-third and an increase in calving interval by one-fourth compared with the previous lactation completed on the farm of origin in a coastal lowland area.

Instrumental insemination failure and facing lower environmental temperatures are probably the origin of these results, while the BCS at the end of the observation period and the data from the feeding trial do not indicate dietary or feeding mistakes. Overall, the distinctive reproduction characteristics of buffaloes, the cooler environmental temperatures and the specificity of the mozzarella production process are the main problems to be faced in an extension of buffalo farming beyond the traditional areas.